GRAIL
What we do Why GRAIL Essays Industry Briefings Book a call

GRAIL · Function papers

What does an AI-augmented data and analytics team look like?

By Johan Grönstedt · Last reviewed

The answer arrives with its meaning attached. This is a nine-page position paper, free.

One email, no newsletter. The PDF opens right here.

A trusted answer becomes a company service, available inside the decision and carrying its definition, sources and uncertainty with it. Analysts still interpret ambiguity. The governed answer does the retrieval, calculation and tracing around them.

The working day

A day in The 2028 analytical rhythm

Most data teams still deliver reports and answer requests. In this rhythm, the trusted answer is available inside the decision with its definition, sources and uncertainty. Analysts interpret ambiguity while the governed answer handles the retrieval, calculation and tracing around them.

The sequence starts with a bounded question and moves through a certified calculation and a trace of its meaning and coverage. An analyst reviews the low-confidence or cross-domain questions. The meeting can then move from reconciling numbers to choosing action.

ASK

A manager asks a bounded question through the company's briefing layer.

CALCULATE

Queries certified models and returns the calculation.

TRACE

Returns the calculation, definition, sources and incomplete coverage.

INTERPRET

An analyst reviews low-confidence answers, questions that cross domains.

DECIDE

The meeting moves from reconciling numbers to choosing action.

The division of work

What runs, and what stays with the person

The agent layer handles the repeatable analytical work around an answer. It drafts queries, visuals, commentary and scheduled briefs against certified measures. For ad hoc work, it clarifies the question, selects approved sources, writes and tests the query, records assumptions and produces a first interpretation. It can also find duplicate measures, draft definitions, identify conflicts and propose tests.

People retain the decisions that establish meaning or turn evidence into action. They approve new measures, consequential narratives, causal claims and executive recommendations. Domain stewards approve meaning, while the central team approves implementation. Engineers retain architecture, security, release authority, schema changes, backfills and source corrections. The agent watches tests and anomalies, traces likely causes, identifies affected assets and drafts remediation.

The same split holds when the work reaches access and data-subject support. The agent maps fields to purposes, assembles evidence, checks retention rules and routes exceptions. The DPO or legal owner decides lawful basis and disputed actions. Across the operating model, the agent retrieves, calculates, compares, traces and drafts an interpretation; the person owns ambiguity, authority and judgment.

ProcessWhat the agent layer doesWhat stays with the person
BI reporting and dashboardsDrafts queries, visuals, commentary, scheduled briefs and threshold alerts against certified measuresApproval of new measures and consequential narratives; the action leaders choose
Ad hoc requests and analysisClarifies the question, selects approved sources, writes and tests the query, records assumptions and produces a first interpretationCausal claims, ambiguous questions and executive recommendations
Metric definitions and governanceFinds duplicate measures, drafts definitions from code and documents, identifies conflicts, proposes tests and shows downstream effectsDomain stewards approve meaning; the central team approves implementation
Data quality and incidentsWatches tests and anomalies, traces likely causes, identifies affected assets and drafts remediationEngineers approve schema changes, backfills and source corrections
Integration and platform operationDrafts mappings, transformation code, tests, documentation and impact analysisEngineers retain architecture, security and release authority

The operating model

Six things GRAIL believes

These six beliefs concern the operating model, not the tools. Together they define who owns meaning, how an answer earns trust and where human judgment enters before evidence becomes action.

The report is no longer the product. The governed answer is.

A dashboard fixes the questions and leaves interpretation to whoever opens it.

The semantic layer is management infrastructure.

Natural-language access over raw tables creates confidence before it creates meaning.

Meaning belongs in the business. Assurance belongs in the centre.

A central team cannot decide what an active customer, net revenue or productive capacity means for every function.

Failed questions are the roadmap.

Every misunderstood term, missing join and stale source shows where the analytical service is incomplete.

Augment, never automate the judgment.

The agent can retrieve, calculate, compare, trace and draft an interpretation.

Start with a decision, then earn the architecture.

The first connection should answer a repeated question that matters across systems.

Get the paper

Read the full nine-page position paper on the AI-augmented data and analytics team. The answer arrives with its definition, sources and uncertainty attached.